Popular Summary

Sensory science is the study of how we measure, analyse, and interpret what our nose, mouth, eyes, ears, and skin perceive ¹. It was born in the 1940s, when the U.S. Army wanted to improve the taste of food rations. Over time, it has become an essential tool for quality control. Its aim is to reduce external and internal biases, such as bad previous experiences with a product or simple mental fatigue. To do that, sensory science standardises products, protocols, and panellists. When biases are reduced, the results become objective, reproducible, and statistically sound ².

Wine is a perfect example of a product whose quality is inseparable from its sensory perception ³. Imagine a glass before you: the clear golden colour, the ripe stone fruit bouquet rising from the glass, a touch of sweetness on the tongue, and a full body wrapping the mouth like a warm hug. All these sensations come together in a simple thought: “this is what I like; this is quality”. Yet, humans are unpredictable instruments. Researcher Isabelle Lesschaeve once compared analytical tools for chemical analysis to humans in sensory analysis ². Instruments are calibrated and checked for accuracy before every measurement. Humans, on the other hand, are far more complex: a bad night’s sleep, a morning coffee, or even a fleeting emotion can shift perception. It is nearly impossible to “calibrate” a person. That is why good sensory science practices exist.

Good sensory science practices

To begin a sensory evaluation, you need a product. In our case, wine – a product that can be seen, smelled, tasted, and felt. Depending on the focus of the analysis, the product must be standardised. For instance, when judging aroma, black glasses are used so that colour doesn’t influence perception. These are the visible tricks. The hidden ones are just as important, like randomising the tasting order to prevent the first or last samples from being overvalued ².

The tasting space also matters. It must have good lighting to judge colour and clarity, a controlled temperature, and no external scents (leave your perfume at home!). Then comes the sensory panel, comprised of people who will evaluate the wines. Panels can consist of consumers, trained assessors, or industry experts. The choice depends on the goal of the panel. If you’re studying consumer preferences, use untrained consumers, and if you’re comparing differences, use trained panellists or experts. Interestingly, studies have shown that trained panellists are more consistent than experts, perhaps because experts often use shortcuts like green reflections indicating unripe fruit, instead of trusting pure perception ³. Panellists shouldn’t handle preparations, present samples, or multitask; they need to stay focused. They also shouldn’t swallow the wine, as alcohol affects perception. Each panellist has an individual spittoon, and between samples, they should rest, breathe, and eat unsalted crackers or drink water to refresh their senses. Trained panels are regularly tested and retrained for accuracy and consistency ².

Sensory science has also been called a philosophy, one that connects consumer, product, chemistry, psychology, and experience ¹. That may sound contradictory, since the field strives to minimise bias, but real-life tasting is never bias-free. People drink wine while watching sunsets, listening to Jimi Hendrix, smelling citronella candles and feeling the fatigue of a long, hot day. Recognising this, sensory science now includes at-home tastings, more representative of how we actually experience wine. These hedonic tests, measuring pleasure and preference, are increasingly valued. The wine industry is rethinking the dominance of experts and trained panels, with new research showing that untrained consumers can provide equally valuable insights ⁴.

Methods

There are many ways to perform sensory evaluation. Here, we focus on two that reveal a wine’s diversity and intensity: the frequency of citation method and RATA (Rate-All-That-Apply). Both are used in the Eco2Wine project to explore whether fermenting with five yeast species at once produces a wine that is more intense and diverse than one fermented with a single species.

First, the experimenter creates a list of descriptors drawn from literature. During training, the panel expands this list. In the actual test, panellists mark every descriptor they perceive, an approach also known as CATA (check-all-that-apply). The goal is to see how often certain attributes are cited. By counting the percentage of panellists mentioning each descriptor per sample, we can visualise relationships between wines and their sensory traits. This method often reveals subtle differences more effectively than classical descriptive analysis (DA), thanks to its broader vocabulary. However, it tends to be less reproducible. In a second step, the descriptor list is refined for intensity assessment using the RATA method. Panellists rate the intensity of each attribute on a 7-point scale. RATA has shown discrimination power comparable to professional DA, even when used with untrained consumers ⁵.

When similar wines are compared, focusing purely on intensity makes evaluation simpler and more precise ⁶. Statistical tools like ANOVA and PCA then identify significant differences and display how wines relate to sensory attributes ⁴. Together, these two methods capture both the breadth and depth of sensory expression, allowing us to see not just what is present in the wine, but how strongly it speaks. Those two tests ensure that the full scope of diversity is captured and that there is a clear evaluation of the intensity of samples.

AI Palate?

Artificial intelligence is transforming sensory science. Imagine a near future where algorithms—not human palates—predict and rate taste experiences with striking precision. Some models already do: trained on non-sensory data, they forecast aromatic profiles from physical, chemical, and physicochemical measurements, replicating the judgments of expert panels. If machines can clone sensory behaviour, we must ask what remains distinctly human in the act of tasting.

Sensory science has always pursued objectivity. Panellists are trained to be neutral, systematic, almost robotic, yet their ultimate purpose is to capture how real people, emotional and inconsistent, perceive flavour. That paradox sits at the heart of the discipline. If wine quality is defined as the absence of faults (a debatable notion), AI could easily detect such deviations. But tasting is never purely chemical. Emotions, context, and personal experience shape every perception. Could AI ever truly taste with the bias, irrational, and beautiful imperfection that defines human experience?

Deeper questions remain. Will dependence on AI redefine what we mean by “quality”? Could it tip the balance between measurable data and lived sensory experience? For AI to be meaningful, it must enrich, not erase, the cultural, emotional, and intuitive dimensions that human tasters bring to the table.

The next frontier may unite both worlds. Researchers are already using virtual reality and multisensory environments to study how mood and setting influence taste. AI could amplify this work, scaling sensory research beyond traditional panels. With enough training data, models might simulate human responses faster, cheaper, and with remarkable precision.

Perhaps one day, AI critics will issue perfectly unbiased wine ratings, replacing arbiters like Jancis Robinson or Robert Parker. But will consumers trust a machine’s palate? The technology evolves at breathtaking speed, but taste, in the end, remains profoundly human. The challenge ahead is not to replace sensory scientists but to redefine their role: to let algorithms refine precision while humans preserve meaning. The story of sensory science is no longer about measuring taste alone; it’s about reimagining how humans and machines might learn to taste together.

Isn’t it weird, that each wine bottle smells so different although it is made from the same fruits? Indeed, grape varieties play an important role in the wine’s aroma profile¹. But there is another small player with a big impact. It’s yeasts. Yeasts are the organisms that transform sugars into alcohol and other molecules. During this process, called fermentation, these different molecules will influence and define the flavour, quality, and almost every aspect of your favourite wine².

In the wine field, yeasts are generally divided into two categories: Saccharomyces and non-Saccharomyces yeasts. This latter category encompasses many different species; among them, we could cite some of the most common ones, present in the vineyard on the grapes and that you may have heard of before, are Hanseniaspora spp., Candida spp., Metschnikowia pulcherrima or Lanchancea thermotolerans³. They all produce their unique cocktail of molecules that give unique flavors to the wine. It is still not fully understood how these microorganisms interactwith each other and what their oenological potential may be⁴.

However, in this article, we will concentrate on the chemical compounds released by these yeasts and the characteristics certain volatile molecules have on wine during fermentation. Those volatile compounds are majorly different from each yeast³.

For example, aromatic compounds like ethyl lactate and lactic acid, formed by L. thermotolerans. respectively add a creamy aspect to the mouthfeel and reduce the acidity and stinginess to create a softer body. M. pulcherrima. contributes to fruity notes such as pineapple and is also associated with other fruit aromas. For pear and apple aromas, their profile is similar to that of S. cerevisaie⁴.

On the other hand, not all chemical compounds contribute positively to wine aroma. Some microorganisms such as Brettanomyces bruxellensis produce some powerful aroma molecules that are associated with a cheesy smell, as well as with stable sometimes⁴.

But then who are we to judge if this is bad ? You may find those aromas a bit particular but quite fine !

In fact, some compounds are seen as a positive contribution if they appear in a small quantity as ethyl guaiacol and ethylphenol formed by B.bruxellensis., which adds a spicy, clove-like aroma for ethyl guaiacol and a little barnyard, leathery for ethylphenol. However, at high levels, the same compounds are considered a defect⁴. As the saying goes, “The dose makes the poison.”⁵. It is the same when considering Candida milleri. While acetic acid, produced as a byproduct of yeast metabolism, can contribute a pleasant freshness at low levels, excessive amounts can lead to vinegar-like aromas, which may negatively impact the wine’s quality⁴. Diacetyl is also an example of something desirable in a low amount but undesirable in a high amount. It gives buttery aromas which can also be unequilibrated in some wines⁴. Getting back to our main yeast, S.cerevisiae, produces byproducts like isoamyl acetate or isoamylol which adds a certain complexity at low concentrations but can be overwhelming at high concentrations⁴.

The balance between desirable and undesirable aromas depends on yeast species, fermentation conditions, winemaking practices, and your taste. What we do, is explore the biodiversity of yeasts and understand their metabolic pathways, which opens possibilities for tailoring wine aromas to enhance complexity and uniqueness. So, if this sparked your interest just continue visiting this site for regular updates on our work.